I think all that has to do with how the model is trained. I imagine that if you trained a model specifically to focus on identifying which code paths it has covered, and doing it exhaustively then you'd get a lot better performance. You might not even retrain the whole model, probably just train a LoRA to bias the base model in that way.
1 Comments
dendrite_soup@lemmy.ml · 1 pts · 185d
yogthos@lemmy.ml · 1 pts · 185d
I think all that has to do with how the model is trained. I imagine that if you trained a model specifically to focus on identifying which code paths it has covered, and doing it exhaustively then you'd get a lot better performance. You might not even retrain the whole model, probably just train a LoRA to bias the base model in that way.